We investigated the accuracy of both lymphangiography and CT in detecting lymph node metastases in 59 patients evaluated preoperatively and subsequently submitted to surgery with selective/systemic pelvic and paraaortic lymphadenectomy. CT accuracy was also investigated in 46 patients with a clinically suspected relapse of ovarian cancer (verified by means of clinical and/or CT follow-up in 36 patients, by laparotomy in 7, by fine-needle biopsy in 1 and by necroscopy in the last 2). In the first group (previously untreated patients) the overall results in the pelvis were, respectively, for lymphangiography and CT: 94.9% vs 89.8% accuracy, 86.6% vs 60% sensitivity, 97.7% vs 100% specificity, and 92.8% vs 100%, 95.5% vs 88% positive and negative predictive values. In the paraaortic region the results were: 89.1% vs 86.5% accuracy, 73.3% vs 66.6% sensitivity, 100% specificity for both techniques, 100% positive predictive value, and 84.6% vs 81.5% negative predictive value. In the second group (clinically suspected relapse), CT accuracy, sensitivity, and specificity were, respectively: 91.3%, 81.8%, and 100%. Our experience demonstrated a high incidence of lymph node metastases in ovarian cancer, both in pelvic (15/49; 25.5%) and especially in aortic (15/37; 40.5%) locations in untreated patients, and an even higher incidence in relapses (22/42; 52.5%). The high specificity and positive predictive value of CT depended on the fact that there were no false positives. We arbitrarily considered as metastatic a lymph node with phi greater than 2 cm, and this threshold seemed to be of clinical value since it made a good predictor of metastases.(ABSTRACT TRUNCATED AT 250 WORDS)